The Pit30M Global Localization Benchmark
We are interested in understanding if retrieval-based localization approaches are good enough in the context of self-driving vehicles.
Towards this goal, we introduce Pit30M, a benchmark for large-scale autonomous vehicle localization comprised of over 30M images and LiDAR sweeps with rich metadata and sub-meter accurate ground truth information.
The dataset is annotated with historical weather and astronomical data, as well as with image and LiDAR semantic segmentation as a proxy measure for occlusion.
Download Link: Coming Soon
We will provide access to our data, including images, LiDAR, ground truth poses, and metadata.
Please stay tuned for more updates.
Localization at City Scale
Pit30M is an image and LiDAR dataset with over 30 million frames, 10 to 100 times larger than any other benchmark.
Pit30M is captured under diverse conditions (i.e., season, weather, time of the day, traffic) and provides sub-meter accurate localization ground truth, as well as rich metadata like weather and semantics.